Perform a histogram test with the metric KLD
This check divides the data into blocks, estimates their probability density functions by histograms and compares them by using the Kullback-Leibler Divergence.
qat_analyse_histogram_test_kld_2d(measurement_vector, blocksize = floor(length(measurement_vector)/20), numofbars = 65, factorofbar = 100)
- The measurement vector (2d array), which should be tested
- Number of elements in the first dimension, which should be used for each block
- Number of bins of the histogram
- Correction factor for non-value bins
The field will be divided into blocks in the first dimension, with a length given by the parameter blocksize. From these blocks histograms are computed and afterwards compared. As a metric for the comparison the Kullback-Leibler Divergence is used. As a result a field is generated, which includes the results of the comparison between every combination of blocks.
Duesterhus, A., Hense, A. (2012) Advanced Information Criterion for Environmental Data Quality Assurance, \_Advances in Science and Research\_, *8*, 99-104.
vec <- array(rnorm(1000), c(100, 20)) vec[51:100, ] <- round(vec[51:100, ]) result <- qat_analyse_histogram_test_kld_2d(vec, 4, 65, 100) qat_plot_histogram_test(result$field, "test_kld_2d", result$blocksize, result$numofbars, result$factorofbar, "kld", result$runs)